Knowledge Organization - users.csc.calpoly.eduusers.csc.calpoly.edu/.../481/W11/Slides/3-Knowledge-Organization.pdf · Franz Kurfess: Knowledge Organization Object Selection what

Computer Science DepartmentCalifornia Polytechnic State University

San Luis Obispo, CA, U.S.A.

Franz J. Kurfess

Knowledge Organization

Some of the material in these slides was developed for a lecture series sponsored by the European Community

under the BPD programwith Vilnius University

as host institution

Acknowledgements

Franz Kurfess: Knowledge Organization

Use and Distribution of these Slides

❖These slides are primarily intended for the students in classes I teach. In some cases, I only make PDF versions publicly available. If you would like to get a copy of the originals (Apple KeyNote or Microsoft PowerPoint), please contact me via email at [email protected]. I hereby grant permission to use them in educational settings. If you do so, it would be nice to send me an email about it. If you’re considering using them in a commercial environment, please contact me first.

3


Overview Knowledge Organization

❖ Motivation, Objectives❖ Chapter Introduction

❖ New topics,Terminology❖ Identification of Knowledge

❖ Object Selection❖ Naming and Description

❖ Categorization❖ Feature-based Categorization❖ Hierarchical Categorization

❖ Knowledge Organization Methods❖ Natural Language❖ Ontologies

❖ Knowledge Organization Tools❖ Editors, visualization tools, automated ontology construction

❖ Examples❖ Important Concepts and Terms❖ Chapter Summary

4


Motivation and Objectives

5


Motivation❖effective utilization of knowledge depends

critically on its organization❖quick access❖ identification of relevant knowledge❖assessment of available knowledge

❖source, reliability, applicability❖knowledge organization is a difficult task, and

requires complementary skills❖expertise in the domain❖knowledge organization skills

❖librarians6


Objectives❖be able to identify the main aspects dealing with

the organization of knowledge❖understand knowledge organization methods❖apply the capabilities of computers to support

knowledge organization❖practice knowledge organization on small bodies

of knowledge❖evaluate frameworks and systems for knowledge

organization

7


Identification of Knowledge❖Object Selection❖Naming and Description

8


Object Selection❖what constitutes a “knowledge object” that is

relevant for a particular task or topic❖physical object, document, concept

❖how can this object be made available in the system

❖example: library❖ is it worth while to add an object to the library’s collection

❖ if so, how can it be integrated❖physical document: book, magazine, report, etc.❖digital document: file, data base, Web page, etc.

9


Naming and Description❖names serve two important roles❖ identification

❖ideally, a unique descriptor that allows the unambiguous selection of the object

❖often an ambiguous descriptor that requires context information ❖ location

❖especially in digital systems, names are used as “address” for an object

❖names, descriptions and relationships to related objects are specified in listings❖dictionary, glossary, thesaurus, ontology, index

10


Knowledge Organization Methods❖Naming and Description Devices❖ index, glossary, dictionary, thesaurus, ontology

❖Natural Language (NL)❖Levels of NL Understanding ❖NL-based indexing

❖Categorization❖Ontologies

11


Naming and Description Devices❖ type❖dictionary, glossary, thesaurus❖ontology❖ index

❖ issues❖arrangement of terms

❖alphabetical, ordered by feature, hierarchical, arbitrary❖purpose

❖explanation, unique identifier, clarification of relationships to other terms, access to further information

12


Dictionary❖ list of words together with a short explanation of their

meanings, or their translations into another language❖ helpful for the identification of knowledge objects,

and their distinction from related ones❖ each entry in a dictionary may be considered an

atomic knowledge object, with the word as name and “entry point”❖may provide cross-references to related knowledge objects

❖ straightforward implementation in digital systems, and easy to integrate into knowledge management systems

13


Glossary❖ list of words, expressions, or technical terms

with an explanation of their meanings❖usually restricted to a particular book, document, activity, or topic

❖provides a clarification of the intended meaning for knowledge objects

❖otherwise similar to dictionary

14


Thesaurus❖collection of synonyms (word sets with identical

or similar meanings)❖ frequently includes words that are related in some other way, e.g. antonyms (opposite meanings), homonyms (same pronunciation or spelling)

❖ identifies and clarifies relationships between words❖not so much an explanation of their meanings

❖may be used to expand search queries in order to find relevant documents that may not contain a particular word

15


Thesaurus Types❖knowledge-based❖ linguistic❖statistical

[Liddy 2000] 16


Knowledge-based Thesaurus❖manually constructed for a specific domain❖ intended for human indexers and searchers❖contains

❖synonyms (“use for” UF)❖more general (“broader term” BT)❖more specific (“narrower” NT)❖otherwise associated words (“related term” RT)

❖example: “data base management systems”❖UF data bases❖BT file organization, management information systems❖NT relational databases ❖RT data base theory, decision support systems

[Liddy 2000] 17


Linguistic Thesaurus❖contains explicit concept hierarchies of several

increasingly specified levels❖words in a group are assumed to be (near-)

synonymous❖selection of the right sense for terms can be difficult

❖examples: Roget’s, WordNet❖often used for query expansion❖synonyms (similar terms)❖hyponyms (more specific terms; subclass)❖hypernyms (more general terms; super-class)

[Liddy 2000] 18


Example 1: Linguistic ThesaurusAbstractRelations

Space Physics Matter Sensation Intellect Vilition Affections

The World

Sensationin General

Touch Taste Smell Sight Hearing

Odor Fragrance Stench Odorless

.1 .9.8.2 .3 .4 .5 .7.6

Incense; joss stick;pastille; frankincense or olibanum; agallock or aloeswood; calambac

[Liddy 2000] 19


[Liddy 2000]

Example 2: WordNet as Linguistic Thesaurus

20


Query Expansion in Search Engines

❖ look up each word in Word Net❖ if the word is found, the set of synonyms from all Synsets

are added to the query representation❖ weigh each added word as 0.8 rather than 1.0❖ results better than plain SMART

❖ variable performance over queries❖ major cause of error: the use of ambiguous words’ Synsets

❖ general thesauri such as Roget’s or WordNet have not been shown conclusively to improve results❖ may sacrifice precision to recall❖ not domain specific❖ not sense disambiguated

[Liddy 2000, Voorhees 1993] 21


Statistical Thesaurus❖ automatic thesaurus construction

❖classes of terms produced are not necessarily synonymous, nor broader, nor narrower

❖rather, words that tend to co-occur with head term❖effectiveness varies considerably depending on technique used

[Liddy 2000] 22


Automatic Thesaurus Construction (Salton)

❖ document collection based❖based on index term similarities❖compute vector similarities for each pair of documents❖ if sufficiently similar, create a thesaurus entry for each term which includes terms from similar document

[Liddy 2000] 23


Sample Automatic Thesaurus Entries

408 dislocation 411 coercive junction demagnetize minority-carrier flux-leakage point contact hysteresis recombine induct transition insensitive409 blast-cooled magnetoresistance heat-flow square-loop heat-transfer threshold410 anneal 412 longitudinal strain transverse

[Liddy 2000] 24


Dynamic Automatic Thesaurus Construction

❖ thesaurus short-cut❖run at query time❖ take all terms in the query into consideration at once ❖ look at frequent words and phrases in the top retrieved documents and add these to the query❖= automatic relevance feedback

[Liddy 2000] 25


Expansion by Association Thesaurus

Query: Impact of the 1986 Immigration LawPhrases retrieved by association in corpus - illegal immigration - statutes - amnesty program - applicability - immigration reform law - seeking amnesty - editorial page article - legal status - naturalization service - immigration act - civil fines - undocumented workers - new immigration law - guest worker - legal immigration - sweeping immigration law - employer sanctions - undocumented aliens

[Liddy 2000] 26


Index❖ listing of words that appear in a set of

documents, together with pointers to the locations where they appear

❖provides a reference to further information concerning a particular word or concept

❖constitutes the basis for computer-based search engines

27


Indexing❖ the process of creating an index from a set of

documents❖one of the core issues in Information Retrieval

❖ manual indexing❖controlled vocabularies, humans go through the

documents❖ semi-automatic❖humans are in control, machines are used for some tasks

❖ automatic❖statistical indexing❖natural-language based indexing

28


Natural Language Methods❖Natural Language Processing❖Natural Language Understanding❖NLP-based Indexing

29

Franz Kurfess: Knowledge Organization[Liddy 2000]

Natural Language Processing❖a range of computational techniques for

analyzing and representing naturally occurring texts❖at one or more levels of linguistic analysis❖ for the purpose of achieving human-like language processing

❖ for a range of tasks or applications

30

Franz Kurfess: Knowledge Organization[Liddy 2000]

NLP-based Indexing❖ the computational process of identifying,

selecting, and extracting useful information from massive volumes of textual data❖ for potential review by indexers❖stand-alone representation of content❖using Natural Language Processing

31


What can NLP Indexing do?❖phrase recognition❖disambiguation❖concept expansion

32


Ontologies❖description❖ “representational promiscuity”❖ontology types❖usage of ontologies❖domain standards and vocabularies

❖ontology development❖development process❖specification languages

33


Categorization❖Hierarchical Categorization❖Feature-based Categorization

34


Hierarchical Categorization❖a set of objects is divided into smaller and

smaller subset, forming a hierarchical structure (tree) with the elementary objects as leaf nodes❖ typically one feature is used to distinguish one category from another

❖often constitutes a relatively stable “backbone” of a knowledge organization scheme

❖re-organization requires a major effort

35


Feature-based Categorization❖objects or documents are assigned to categories

according to commonalties in specific features❖can be used to dynamically group objects into

categories that are of interest for a particular task or purpose❖re-organization is easy with computer support

36


Ontology❖examines the relationships between words, and

the corresponding concepts and objects❖ in practice, it often combines aspects of thesaurus and dictionary

❖ frequently uses a graph-based visual representation to indicated relationships between words

❖used to identify and specify a vocabulary for a particular subject or task

37


The Notion of Ontology❖ontology

explicit specification of a shared conceptualization that holds in a particular context

❖captures a viewpoint on a domain: ❖ taxonomies of species❖physical, functional, & behavioral system descriptions❖ task perspective: instruction, planning

[Schreiber 2000] 38

Franz Kurfess: Knowledge Organization[Schreiber 2000]

Ontology Types❖domain-oriented

❖domain-specific ❖medicine => cardiology => rhythm disorders❖ traffic light control system

❖domain generalizations ❖components, organs, documents

❖ task-oriented❖task-specific

❖configuration design, instruction, planning❖task generalizations

❖problems solving, e.g. upml

❖generic ontologies ❖ “top-level categories”❖units and dimensions

39


Using Ontologies❖ ontologies needed for an application are typically a

mix of several ontology types❖ technical manuals

❖device terminology: traffic light system❖document structure and syntax❖ instructional categories

❖e-commerce❖ raises need for❖modularization❖ integration

❖ import/export❖mapping

[Schreiber 2000] 40


Domain Standards and Vocabularies As Ontologies

❖ example: Art and Architecture Thesaurus (AAT)❖ contains ontological information

❖ AAT: structure of the hierarchy❖ structure needs to be “extracted”

❖ not explicit❖ can be made available as an ontology

❖ with help of some mapping formalism❖ lists of domain terms are sometimes also called “ontologies”

❖ implies a weaker notion of ontology❖ scope typically much broader than a specific application domain❖ example: domain glossaries, wordnet❖ contain some meta information: hyponyms, synonyms, text

[Schreiber 2000] 41


Ontology Development

Scott Patterson, CS8350

Kietz, Maedche, Voltz; A Method for Semi-Automatic Ontology acquisition from a Corporate Intranet

Maedche & Staab; Ontology Learning for the Semantic Web

DomainOntology

Extract

Import/Reuse

Prune

Refine

Select Sources

Concept Learning

Relation learning

Evaluation

42


Ontology Specification❖many different languages❖KIF❖Ontolingua❖Express ❖LOOM❖UML❖XML to the rescue: Web Ontology Language (OWL)

❖common basis❖class (concept)❖subclass with inheritance❖relation (slot)

[Schreiber 2000] 43


Knowledge Organization Examples

❖ad-hoc via diagrams❖concept-form-referent triangle❖ontology mind map❖comparison on knowledge organization methods❖ taxonomy, thesaurus, topic map, ontology

❖examples of ontologies

44


Knowledge Organization Example (ad-hoc diagram)

http://keg.cs.tsinghua.edu.cn/persons/tj/Reports/Pswmp-Jie-Tang.ppt45


^

Communication Principle

ReferentForm Stands for

refers toevokes

Concept

“Jaguar“

[Odwen, Richards, 1923]

[Hotho, Sure, 2003]

46


Views on OntologiesFront-End

Back-End

TopicMaps

Extended ER-Models

Thesauri

Predicate Logic

Semantic Networks

Taxonomies

Ontologies

Navigation

Queries

Sharing of Knowledge

Information Retrieval

Query Expansion

MediationReasoning

Consistency CheckingEAI

[Hotho, Sure, 2003] 47


Extending Taxonomies to Ontologies

❖ Taxonomy❖strict hierarchy

❖ Thesaurus❖hierarchy plus synonyms and other relations between words

❖ Topic Map❖additional relations between concepts

❖across the hierarchy❖properties of concepts

❖ Ontology❖ rules specifying the structure of the concept space❖ instances of concepts

48


Object

Person Topic Document

ResearcherStudent Semantics

OntologyDoctoral Student

Taxonomy: Segmentation, classification and ordering of elements into a classification system according to their relationships between each other

PhD Student F-Logic

Menu

[Hotho, Sure, 2003]

Taxonomy

49


Object



PhD StudentDoktoral Student

•Terminology for specific domain•Graph with primitives, 2 fixed relationships (similar, synonym), sometimes

additional relationships (antonym, homonym, ...) •originated from bibliography

similarsynonym

OntologyF-Logic

Menu

Thesaurus



Object




knows described_in

writes

AffiliationTel

• Topics (nodes), relationships and occurrences (to documents)• ISO-Standard• typically for navigation and visualization

OntologyF-Logic

similarsynonym

Menu

Topic Map



OntologyF-Logic

similar


Object


Tel

Semantics

knows described_in

writes

Affiliationdescribed_in is_about

knowsP writes D is_about T P T

DT T D

Rules

subTopicOf

• Representation Language: Predicate Logic (F-Logic)• Standards: RDF(S); OWL

ResearcherStudent

instance_of

is_a

is_a

is_a

Affiliation

Affiliation

York Sure

AIFB+49 721 608 6592

Ontology



Knowledge Organization

Examples

53


Cyc Knowledge Base Structure

54

Follow the link below for an interactive version that shows more information about the categories (requires JavaScript, and may not work in all browsers):http://www.cyc.com/cyc/images/cyc/technology/whatiscyc_dir/whatdoescycknow


OntoWeb.org

Portal Generation Navigation

Query/SerachContent

Integration Collect metadata from participating partners

Annotation [Hotho, Sure, 2003] 55


Art & Architecture Thesaurus

used forindexing stolen art objects in Europeanpolice databases

[Schreiber 2000] 56


AAT Ontologydescriptionuniverse

descriptiondimension

descriptor

value set

value

descriptorvalue

object

object type object class

classconstraint

has feature

descriptorvalue set

in dimension

instance of

class of

hasdescriptor

1+

1+

1+

1+

1+

1+

[Schreiber 2000] 57


ARNET Miner 1

58


ARNET Miner 2❖ `

59


Top-level Categories:Many Different Proposals

Chandrasekaran et al. (1999)

[Schreiber 2000] 60


Important Concepts and Terms

61

❖ automated reasoning❖ belief network❖ cognitive science❖ computer science❖ deduction❖ frame❖ human problem solving❖ inference❖ intelligence❖ knowledge acquisition❖ knowledge representation❖ linguistics❖ logic❖ machine learning❖ natural language❖ ontology❖ ontological commitment❖ predicate logic❖ probabilistic reasoning❖ propositional logic❖ psychology❖ rational agent


Summary

62

Knowledge Organization - users.csc.calpoly.eduusers.csc.calpoly.edu/.../481/W11/Slides/3-Knowledge-Organization.pdf · Franz Kurfess: Knowledge Organization Object Selection what

Documents